Every few years, we get new interfaces. Normally, they’re spread out a bit. USB 2.0 comes out, then a new SATA version and later a new PCI Express revision. Lately, though, the trickle of new interfaces has become a deluge, and keeping up with all of them can be mind-numbing – not to mention hard on your credit card.
Let’s take a look at both recently arrived interfaces and those on the near term horizon. We’ll also try to figure out when it makes sense to upgrade or move to the new connection or wait for something better.
It’s been a long time coming, and it’s finally here – sort of. A growing list of peripherals – particularly external hard drives like Seagate’s Black Armor 110 – are shipping with USB 3.0 support. The downside: most off-the-shelf PC systems don’t come with USB 3.0. The major chipset makers, AMD and Intel, haven’t integrated USB 3.0 yet, though we hear that AMD may be readying a south bridge with support for the new standard. Intel likely won’t be integrating native support into its core logic for quite a few months yet.
The good news is that most motherboard makers are aggressively adding discrete USB 3.0 controllers to their retail products. Gigabyte and Asus, for example, are even adding USB 3.0 controller chips to modestly priced micro ATX boards, like the Asus P7H55D-M Evo and Gigabyte’s GA-H57M-USB3. So if you’re building your own PC, adding USB 3 support is fairly easy. However, if you’re buying a PC from one of the major PC vendors, you may have to add a USB 3.0 expansion cards. The peripheral vendors are stepping up to the plate – the aforementioned Black Armor 110 comes with an ExpressCard USB 3.0 adapter – though that won’t help much if your desktop PC lacks a USB 3 port.
USB 3, aka “SuperSpeed USB”, is likely to replace eSATA for connecting external storage. So it’s probably worth investing in an expansion card to add USB 3.0 to take advantage of the increasing number of external hard drives supporting the standard.
I’ve seen this called “SATA 3” in some motherboard manuals, but the correct term is SATA I/O 6gbps. This is the latest update to SATA. As with USB 3.0, no chipset supports the new standard natively. (Intel, oh, Intel, why do you let me down so?)
Why a faster SATA if hard drives are barely hitting up against the 3gbps barrier – and even then, only in the outer tracks? The answer, of course, is SSDs. The new generation of SSDs will saturate 3gbps quiet readily, so a faster SATA interface is needed. If you’re still using rotating magnetic platters because of cost or capacity constraints, you can wait.
Motherboard makers are mostly integrating SATA 6gbps into high end boards, which makes sense. Professional workstations and servers may benefit, particularly if you’re building multi-drive striped arrays and need maximum throughput. But average users may not really see much gain.
Let’s face it: DVI is a clunky, aging interface. It’s clunky because it uses this… frickin’ huge connector, at least by today’s standards. Plus, you have to keep track of multiple different types – DVI-I, DVI-D, dual link DVI. This results in cable management hell.
Analog VGA… er, the less said, the better. Just kill it already.
When DisplayPort first came out, I was skeptical. After all, we had HDMI. But as I’ve used it more, I’m sold. The connector is compact, since it’s a serial interface. It’s oriented towards the high bandwidth needs of PC users. Since the connector takes up so little real estate, it’s easy for graphics card makers to integrate multiple DisplayPort connectors onto the back of a video card. (So… AMD… when will we see that card with six DisplayPort connectors?)
There are downsides if you’re moving from DVI to DisplayPort. Most GPUs only support two active digital connectors, usually because they only have so many timing blocks built into the GPU. So if you want to connect more than two monitors, and only have DVI-equipped displays, you’ll need at least one active DisplayPort to DVI adapter, which is pretty pricey. Expect to pay a hundred buckazoids or more.
DisplayPort assumes that the monitor will handle all the timing signals, which simplifies circuitry on the graphics card side of the equation. This is why a card can theoretically support more than two monitors – it doesn’t need all that pesky timing circuitry on board the card.
We’re starting to see quite a number of DisplayPort equipped PC monitors arriving, which is a good thing. So if you’re buying a new monitor, look for DisplayPort. It’s definitely worth the move. But if your current monitor setup is working for you, don’t rush out to upgrade. Do it when the timing is right for you.
Will the endless stream of HDMI versions please just stop already?
Okay, I don’t really mean that. The consumer electronics companies’ hunger for pumping out new features almost requires new HDMI updates. The good news is that HDMI seems to be a more significant update than a mere point revision number might suggest, increasing the bandwidth substantially to support higher refresh rates at HD resolution – something needed for stereoscopic 3D HDTV, which is coming sooner than you think.
HDMI adds some other cool stuff, too. One of the side effects of the increased bandwidth is support for higher resolutions, including 4K support (3840 x 2160 pixels.) Also added in is an HDMI Ethernet channel, which allows easier networking (though the spec only allows for fast Ethernet over HDMI rather than gigabit speeds) and IP support. Bigger color spaces, smaller connectors and in-car features have also been added.
I would hope, however, that we see an end to HDMI updates for at least a couple of years.
(You in the back row: shut up about 4K 3D. Just shut up!)
Just when we’ve gotten used to PCIe 2.0 graphics cards and motherboards, another version of the now-venerable standard is imminent. PCI Express 3.0 increases the bandwidth from 5 gigatransfers per second to 8 GT/sec. But it’s actually faster than the number might indicate. PCIe 2.0 used what’s known as an 8b/10b encoding, which meant 20% of the bits were overhead. PCIe 3.0 will use 128b/130b encoding, which reduces overhead to a mere 1.5%. So the net effect is to roughly double the throughput.
PCI Express 3.0 will also likely deliver power beyond the 75W the current slot can handle, but the standard still hasn’t been finalized. Rumors abound, though, with numbers like 300W bandied about. The actual PCIe 3.0 final spec won’t see the light of day until at least Q2 of 2010. We may see products by year’s end, but that will be a stretch.
Light Peak, as the name implies, is an optical interface designed to deliver very high bandwidths. Light Peak’s feasibility is made possible by Corning’s ClearCurve optical cable technology, which can bend around short curves without cracking or breaking. ClearCurve cabling is suitable for peripheral connections ranging from very small (handheld devices) to high speed server interconnects. First implementations are supposed to support speeds up to 10 gigabits per second, with 100gb/sec theoretically possible.
Developed mainly by Intel, the goal for Light Peak is to completely replace existing copper-based standards, including USB, FireWire, SATA, HDMI and others. Imagine that: a single physical infrastructure replacing a host of different connectors, capable of supporting multiple protocols as needed.
Of course, if Light Peak delivers on its promises, you’ll need to ride the upgrade cycle again. Keep that credit card handy.